Newest papers

Benchmarking Large Language Models for Detecting Satirical Code-Mixed Sentences

Răzvan-Alexandru Smădu, Ioana Buhnila, Daniela Tănase, Dumitru-Clementin Cercel, Florin Pop, Mihaela-Claudia Cercel · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.29907 · cited by 0 works (OpenAlex) · open-access PDF

Portuguese

As tarefas de processamento de linguagem natural, como a detecção de satira, dependem fortemente de sinais culturais e sintáticos que os grandes modelos de linguagem lutam para capturar. Neste artigo, investigamos quatro métodos (ou seja, substituição aleatória, tradução de frases, análise de dependência e modelos de idiomas grandes) para gerar sentenças sintéticas de código misto em romeno com inglês, alemão, francês e italiano. Começamos com um conjunto de dados existente para a detecção de satira, compilado a partir de artigos de notícias romenas e anotado no nível de sentença por humanos, e aumentamos com mais de 600 exemplares, nomeadamente CS-SeRoRoRo. Fornecemos uma análise profunda do processo de geração de dados. Nós fornecemos uma análise aprofundada do

Machine translation (Meta M2M100); the English original follows.

Abstract

Code-switching is a natural linguistic phenomenon, yet its impact on high-level reasoning tasks remains largely unexplored in under-resourced languages like Romanian. Natural language processing tasks, such as satire detection, rely heavily on nuanced cultural and syntactic cues that large language models struggle to capture. In this paper, we investigate four methods (i.e., random replacement, phrase translation, dependency parsing, and large language models) for generating synthetic code-mixed Romanian sentences with English, German, French, and Italian. We start with an existing dataset for satire detection, compiled from Romanian news articles and annotated at the sentence level by humans, and augment it with over 600k samples, namely CS-SeLeRoSa. We provide an in-depth analysis of the dataset generation process. Then, we evaluate several multilingual large language models across zero-shot, chain-of-thought, and fine-tuning regimes on both the synthetic corpus and a human-adjudicated gold benchmark, namely CS-SeLeRoSa-Gold. While older open models suffer substantial degradation (dropping to 37%-55% macro-F1), newer models like Gemma 4 31B achieve 68.1% zero-shot, rivaling frontier models like GPT-5 mini (69.5%). Furthermore, fine-tuning significantly reduces formatting errors and improves the results, increasing Gemma 4 31B to 77.5% on CS-SeLeRoSa and 72.6% on the gold standard, outperforming frontier zero-shot baselines.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Natural language processing, Artificial intelligence, Language model, Sentence, Phrase

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange